Re: experience w/ UUID's for surrogate keys across entire schema?

  • From: Sayan Malakshinov <xt.and.r@xxxxxxxxx>
  • To: dmarc-noreply@xxxxxxxxxxxxx
  • Date: Fri, 14 Sep 2018 15:46:49 +0300

Hi Chris,

Of course, you can extend option 2 as multicolumn (SITE_ID,UUID) and use
partitioning by SITE_ID to avoid some performance problems, but anyway
because of the nature of UUID all inserts will require split operations on
indexes, so sequences usually are better choice. It's much easier to solve
problems with concurrency for right block of index (for example, using hash
partitioning) than deal with constant splits


-- 
Best regards,
Sayan Malakshinov
http://orasql.org

Other related posts: